Open source is not free (as in beer) ...

Last week, I took the time to ridicule a post at the PBBI blog on open source, which really boiled down to a critique of the very flimsy open source argument “open source is free (as in beer) so it’s a budget panacea!”. I hope not many open source advocates are retailing that one, but I’m sure one or two of us still are.

In my idealistic younger days I read a lot of Chomsky, and one of the general principles I took away from him is the idea that as citizens we have a duty to critique and criticize our own governments first and foremost. Beating our chests about the behaviour of some despot on the other side of an ocean might be cathartic, but if we’re really interested in improving the commonweal it is our job as citizens of democratic societies to make our own governments better first. (Like, why does Canada sell asbestos as a building supply to third world countries when back home we recognize that it’s too dangerous for our own buildings?)

Anyhow, in that spirit, it is really my responsibility, not PBBI’s, to point out the disingenuousness of “open source is free (as in beer)” as a slogan.

So, first the counterargument.

I think most folks recognize that implementation costs will wash out software acquisition costs for any moderately complex system build. But we still, in our lizard brains, feel like the acquisition cost of proprietary software has to be a systemic problem. Perhaps because it’s so painful at a personal level to whip out the credit card and plunk down a few hundred bucks for something like Microsoft Office.

However, when you examine the numbers, it gets very clear very fast that service costs are an order of magnitude larger than software acquisition costs. The BC Public Accounts tell the story for my home province: For the past ten years, ESRI has taken in between $1M and $2M a year pretty consistently selling GIS software. Over those same years, the consulting companies in town I would call “geospatial companies” have in aggregate billed about 10 times that annual figure providing services.

Services rule! The profit margins in services are notoriously lower than for software, and I’m sure ESRI made more pure profit on their $1M per year. But the bulk size of service revenues more than compensates. That’s why IBM and HP are growing into services. Because there’s so much room to grow in there.

Second, the refinement.

While “open source is free (as in beer)!” can be oversold, it is not actually untrue. Open source does actually have a price-tag of $0, and proprietary has a price-tag of > $0. So how to refine the slogan to not over-emphasize price? I have been using a modification which I hope focusses people more clearly on where the price can be important.

“Open source has a zero dollar capital cost.”

So, given that the cost of services out-weighs software for most system building, you might not care about software acquisition cost if you’re deploying the finished product once. But if you’re deploying a system across 10 or 100 nodes, then the capital cost of deployment might be an extremely important economic variable.

In some cases (cough Oracle cough) the cost of just one node can be enough to push people over the edge on initial capital cost. But most proprietary software is more moderately priced, and it takes more nodes before the pure price differential is worth feeding into a comparison exercise.


I lay out more non-monetary reasons for bringing open source to the management table in my talk A Manager’s Guide to Open Source.

Proprietary software is not the future you think it is...

OK, so this morning James Fee tweeted this nugget of late 90’s “open minded thinking about open source” from a proprietary vendor.

And if you think I can leave that alone… you just don’t know me that well!

I’d go through and work on the arguments point-by-point, but there’s hardly any argument there. It’s almost all innuendo and unsupported assertion. So I just re-wrote it from the opposite point of view largely using search-and-replace. It holds up pretty well, because there’s no content, just bluster. My favourite part is the paragraph on PostGIS, which I left unchanged, since it’s actually a pretty powerful argument against proprietary software (your pro-proprietary argument is that Oracle is going to kick your ass if you use open source? awesome!)

Enjoy.


Proprietary software is not the future you think it is…

Read those words carefully. Read what they are, and understand what they are not. It could mean “open source zealot dismisses proprietary software”. It doesn’t. What it means is that while many preach the virtues of proprietary software, meaning supported, reliable, and guaranteed uptime with no technical expertise required – I do not subscribe. Closed platforms, be it maps or software, are fantastic and I believe a part of the IT industry as far as we can see into the future. OpenGeo, for example, uses proprietary software in its own products so the likes of the OpenGeo Suite connect easily to an ESRI ArcSDE server. But let me give you a couple of examples where I am coming from.

I saw a presentation a couple of years ago from a older gentleman – about my age – who had a 30 minute slot at a conference. For 25 minutes he exalted the virtues of his product, and how he can take this piece of software, and solve all kinds of business problems with just the touch of a button and you – yes you – can touch that button too. But he was also pushing his wares, and spent the final five minutes telling you how his product would be even more useful if you also bought all this other products. So the end result in that case was: this tool is great, but to be actually useful you have to buy all the other stuff. Come again? You mean its good, but I can only unlock the value by standardizing on your whole platform? But you just spent 25 minutes telling me how great this new stuff is and now I’m still being asked to open my wallet even more? Heckling ensued from the peanut gallery.

Let’s take another scenario. You have decided to invest in an standard vendor solution. To put this all together you decide to employ one or two low-end hacks from a technical training farm. Over the following 12 months these guys struggle (because they aren’t very bright) but in the end you have something that works, and only requires a nightly reboot.

But a year or two down the line, your management decides they need the system to support ten times as many users, and provide 99.9% up-time. Maybe you don’t have the budget to license that many cores. Now you’ve got to re-architect and re-write the system to scale and not require reboots anymore. And there’s a problem. Your low-end hacks can’t figure out how to do anything that’s not in the “introduction” section of the manual or can’t be done with the GUI. Given the choice between learning how to script a solution and spending 5 hours clicking the same buttons over and over, they choose the latter. Any why not? They are on salary, after all. You can see where this is going.

The point of course, is that proprietary software is not 100% reliable or scalable with zero effort. And proprietary software may not even be the best engineered solution. I am not asking you to ignore proprietary software and only use open source, because that in some ways puts you back to square one. Proprietary software has credible value to both system integrators providing sustainable solutions and individual organizations implementing systems.

Take Oracle as an example. It has been the industry standard for years, and is in many instances a credible alternative to the likes of PostgreSQL. But are you going to pay more or less for that Oracle DBA over the PostgreSQL DBA? Assuming it’s the same, or perhaps a little more for a specialist, then we’re into pure software licensing. I can guarantee Oracle aren’t about to roll over because of pricing. If a company has a site agreement with Oracle are you free to pick up another database? Even if it is free – or free to download? As some people say “spatial is special” we are increasingly seeing that IT departments disagree. Spatial is increasingly not viewed as special and won’t be treated as such.

I’ve already seen firsthand, organizations that have tried and given up on building a sustainable product to sell based entirely on proprietary software, and organizations do a complete U-turn on what they were told were “off-the-shelf” systems because of spiralling costs and concerns over maintenance. It’s not to say you shouldn’t take that path, but it is to say, it’s not easy, and in the long run it might not even be possible. The crux behind all of this is money. If we all had loads it wouldn’t matter so much. We’d buy whatever system we wanted, and lots of consultants to wire the black boxes together. But now more than ever we don’t have money. If an offer seems too good to be true, then it probably is. Proprietary software will continue to evolve and grow and I’ve no doubt more and more individuals and organizations will participate. Some organizations will use proprietary software for the press releases and “president’s leadership” awards, others for the free lunches with sales reps. I still believe we are looking for products which give us the best of both worlds. Extensibility where we are comfortable, but somehow reducing the risk and ongoing support to minimal predictable cost.

A couple of weeks ago I said a more informed decision is likely to be a better one. Next time someone tells you proprietary software is cheaper overall and more reliable, have a little think about it. Use what is right, not what someone else says might be cheap.

Paul R

PS – This blog is using blogger – we use it because it’s a great tool for the job, not just because it’s closed and proprietary!

Data vs Reports

I’m sure I’ll have to spend a lot more time trolling the DataBC site before I have discovered all the jewels therein amongst the dross, but one thing I noticed while searching for “birth” was a large number of entries like this entry from Vital Statistics: “Table 25 - Infant Mortality by Gestational Age and Birth Weight, British Columbia, 2005”

Is this data? Kinda. There’s numbers. It’s an Excel sheet. But let’s get real here, this is a report. It’s a summary. That’s why Vital Statistics has hundreds and hundreds of “data set” entries like this in the catalogue instead of the number they should have: ONE.

There is one database of births and deaths in the province. Strip out the names and addresses, leave the causes, dates, ages, postal codes, and release THAT please, instead of the hundreds of year-by-year (not even longitudinal summaries! arg!) summaries.

Guh, Goh!

Remember Google Wave? Maybe not, but you might remember the hype. The 1 hour video introduction, the breathlessly handed out friend-of-a-friend invitations. Google Buzz, similarly. Google Health. Orkut. Knol. The list goes on.

The upside of trying lots of things is that you sometimes catch a winner, I guess. Maps and location were winners, but they have a natural synergy with the search / advertising business.

The downside is that Google has burned up all it’s cachet. When 80% of the products you launch bomb, people eventually assume everything you launch is going to bomb. Which, in the self-levitating bootstrapping world of network services and network effects means they have already bombed. Google+, we hardly knew thee.

Visualizing Progress

Data visualization gets (from me) a bad rap, because so often it is used as a way to make fundamentally unsexy things just sexy enough for management to buy, without reference to whether the item in question is, at core, useful or effective.

Caveat aside, this graph from the GeoTools JIRA side made my hair stand on end.

Yikes, do you close more tickets than get opened? There’s a visualization that lights a fire under my butt!

Even when I’m in full ticket-hunting-mode I feel like I sometimes am not making forward progress, since Regina gets into bug-finding-mode and starts opening them up on me faster. And the last year of adding features to PostGIS (raster, topology) while altering core structures (serialization, indexes) has created an avalanche of tickets. More than fits on a page. Time to get closing!